CISI

CISI predicts cross-interaction and self-interaction tendencies of monoclonal antibodies from tripeptide composition to inform development and quality assessment.

Key Features:

• Predictive modeling: Predicts cross-interaction and self-interaction propensities of monoclonal antibodies using tripeptide composition as input.
• Input data: Uses tripeptide composition derived from antibody sequences as the feature set for prediction.
• Performance metrics: Accuracy 88.20%; Sensitivity 90.22%; Specificity 86.05%; Matthews Correlation Coefficient (MCC) 0.78; Area Under the ROC Curve (AUC) 0.96.
• Validation methodology: Model performance was evaluated using leave-one-out cross-validation.

Scientific Applications:

• Drug Development Efficiency: Predicts potential cross- or self-interactions early to help mitigate delays and reduce costs in monoclonal antibody development.
• Quality Control: Identifies antibodies with interaction tendencies that could affect therapeutic efficacy or safety.

Methodology:

Analysis of tripeptide composition to predict interaction tendencies, with model performance validated by leave-one-out cross-validation and grounded in empirical assay data from poly-specificity reagent, cross-interaction chromatography, biolayer interferometry, and affinity-capture self-interaction nanoparticle spectroscopy.